Association of Both High and Low Left Ventricular Ejection Fraction With Increased Risk After Coronary Artery Bypass Grafting

Abstract

Background.

While reduced left ventricular ejection fraction (LVEF) is a known risk factor for complications after coronary artery bypass grafting (CABG), the relevance of higher LVEF values has not been established. Currently, most risk stratification tools consider LVEF values above a certain point as normal. However, since this does not account for insufficient ventricular filling or increased adrenergic tone, higher values may have clinical significance. To improve our understanding of this situation, we investigated the relationship of preoperative LVEF values with short- and long-term outcomes after CABG using a strategy that allowed for the identification of nonlinear relationships. We hypothesised that both higher and lower values are independently associated with increased postoperative complications and death in this population.

Methods.

We performed a single-centre retrospective cohort study of patients undergoing isolated CABG surgery. All patients had a preoperative measurement of their LVEF. Surgery involving mitral valve repair was excluded in order to eliminate the impact of mitral regurgitation. The primary outcome was long-term mortality; secondary outcomes included atrial fibrillation, operative mortality, and a composite outcome including any postoperative adverse event. Fractional polynomial equations were used to model the relationship between LVEF and outcomes so we could account for nonlinear relationships if present. Adjustments for confounders were made using multivariable logistic regression and Cox models.

Results.

A total of 7,932 subjects were included in the study. After adjusting for patient and surgical characteristics, LVEF remained associated with the primary outcome as well as the composite outcome of any postoperative adverse event. Both these relationships were best described by a J-shaped curve given that higher LVEF values were associated with increased risk, albeit not as high has lower values. Regarding long-term mortality, individuals with a preoperative LVEF of 60% demonstrated the longest survival. A statistically significant relationship was not found between LVEF and operative mortality or atrial fibrillation after adjustment for confounders.

Conclusions.

Higher preoperative LVEF values may be associated with increased risk for patients undergoing CABG surgery. Future studies are needed to better characterise this phenotype.

Introduction

Reduced systolic function, typically quantified as the left ventricular ejection fraction (LVEF), is a known risk factor for postoperative complications among patients undergoing both cardiac and noncardiac surgery. For those undergoing coronary artery bypass grafting (CABG), reduced LVEF is most commonly associated with ischaemic cardiomyopathy and is known to increase the risk of postoperative adverse events [1]. Specifically, the incidence of low cardiac output syndrome, respiratory failure, acute kidney injury, atrial fibrillation, pneumonia, sepsis, stroke, and haemorrhage are all increased in this population [2–9]. Some, but not all studies, have also found associations between lower LVEF and increased late mortality [10,11].

While the relationship between reduced LVEF and CABG outcomes is well described, few studies have investigated the importance of higher values. This is an important knowledge gap since recent studies have described decreased survival in the general population for individuals with an LVEF above or below 60–65% [12]. Consistent with this, we recently described an association between higher LVEF values and postoperative adverse events in patients undergoing noncardiac surgery, with the lowest risk occurring with preoperative values between 58 and 62% [13]. Therefore, patients with higher LVEF values may represent an important clinical phenotype.

In order to determine if a similar relationship exists for patients undergoing cardiac surgery, we studied the shape of the relationship between preoperative LVEF and postoperative adverse events in patients undergoing CABG. We hypothesised that both lower and higher preoperative LVEF values are associated with increased postoperative mortality based on our previous findings in patients undergoing noncardiac surgery.

Patients and Methods

Study Population

This study was approved by the Institutional Review Board at Mercy St. Vincent Medical Center, Toledo, OH which waived informed consent. This study followed the Strengthening of Reporting of Observational Studies in Epidemiology guidelines. All patients who underwent isolated CABG surgery between 13 January 1995 and 3 September 2012 were considered for inclusion. Surgeries after 2012 were not included because patients were not entered into the database after this time. Individuals without a preoperative LVEF measurement were excluded. The LVEF data in the dataset were primarily obtained from a preoperative ventriculogram. All data were obtained from the de-identified version of the institution’s Society of Thoracic Surgeons Adult Cardiac Surgery database using the definitions in place at the time of the index surgery.

Outcomes

For our primary outcome of late mortality, we obtained post-discharge deaths from institutional follow-up and verified from recurrent queries of: 1) US Social Security Death Index (http://ssdi.genealogy.rootsweb.com) (last checked November 2011 after which this resource was no longer consider a valid research tool) and 2) For Ohio residents from the 1994–2013 Ohio Death Index. The US Social Security Death Index was used through November of 2011 given its availability and accuracy [14]. After December of 2011, this source of information became less accurate [15], prompting us to obtain mortality data after this time from the Ohio Death Index. For the late follow-up analysis (Cox hazard models), non-Ohio residents were censored at Nov 2011 since it was unlikely that they would be included in the Ohio Death Index.

Secondary outcomes were obtained from the institutional Society of Thoracic Surgeons Adult Cardiac Surgery database and included operative mortality, new onset atrial fibrillation, and a composite of all postoperative complications including death, reoperation, new onset atrial fibrillation, acute kidney injury (AKI), respiratory failure, and pneumonia. Operative, or early, mortality was defined as death within 30 days of surgery or prior to hospital discharge or any death occurring prior to hospital discharge if the patient remained hospitalised beyond 30 days of surgery. This definition was selected to be consistent with the definition of operative mortality used by the Society of Thoracic Surgeons [16]. New onset atrial fibrillation was additionally studied as a separate secondary outcome, since it is the most common postoperative complication.

Covariates

Patient and surgical factors were analysed as covariates. These included demographic information, the presence of comorbidities, use of certain medications, and processes of care.

Statistical Analysis

Continuous variables were summarised using the mean ± standard deviation (SD) and categorical variables were summarised using the frequency and percent. Comparison of continuous and categorical variables between groups were made using Student’s t-test or chi square test, respectively. When data were missing for categorical variables, an extra categorical level was created since data are often not missing at random [17]. Missing data for continuous variables were handled via a complete case analysis.

In order to detect nonlinear relationships between LVEF and postoperative adverse events, fractional polynomial equations weighted for the frequency of each LVEF were used to transform LVEF values to a fitted curve [18–20]. Separate transforms were done for each outcome. The coefficients of the fractional polynomials were found using SOLVER in Excel (Microsoft, Redmond, WA, USA).

To determine the factors independently associated with perioperative mortality, complications, and new onset atrial fibrillation, we entered all demographics, comorbidities, intraoperative processes of care, and the transformed LVEF into three separate binary logistic regressions—one for each of the outcomes. Model selection was done with forward selection, p<0.1 to enter and <0.05 to stay. Similarly, for late mortality, we determined the factors independently associated with overall mortality by entering all factors and the transformed LVEF into a Cox model after examining the Kaplan-Meier curve to verify the proportional hazard assumption. Factors with odds ratios or hazard ratio with p<0.05 were considered statistically significant. If the transformed LVEF was in the final models, it was then back-transformed to determine the odds ratio and hazard ratios of LVEF and presented graphically as risks. As the emergent-salvage group was small, it was combined with the emergent group for all multivariable analyses. All statistics were done using SPSS 24 (IBM Corp., Armonk, NY, USA).

Results

Removing the 477 (6%) individuals who were missing a preoperative LVEF resulted in a final study population of 7,932 individuals. Patients with and without a preoperative LVEF measurement were similar in all characteristics except for surgical status (p<0.001) and race (p<0.001). The proportion of patients with and without this measurement were 6.8% vs. 18% for emergent or emergency salvage status, 60% vs. 47% for urgent status, and 33% vs 36% for elective status, respectively. Regarding race, the proportion of patients with and without a preoperative LVEF value were 77% vs. 91% for White race, 4.6% vs. 3.7% for Black race, 2.7% vs 2.1% for Other race, and 16% vs. 3.4% for Unknown race, respectively.

Of patients included in the analysis, most were male, White, and covered by Medicare or private insurance. Prior cardiac procedures were common with 2,571 subjects (32%) having undergone a prior intervention. Of these, 384 (15%) had undergone a previous CABG surgery, and 97 (4%) had undergone a different type of cardiac surgery. Only 2,626 (33%) patients were undergoing an elective surgery. Subjects received an average of 3.2±1.0 bypass grafts. Ejection fractions ranged from 5% to 89% (Figure 1). Overall, the mean ejection fraction was slightly depressed, 48±11%. Forty-four per cent (44%) (n=3,454) had an LVEF >55%. Moderately to severely reduced LVEF values were present in 1,424 subjects (18%), 127 (1.6%) had ejection fractions >65% (Table 1).

Figure 1.

Histogram showing the distribution of preoperative LVEF values included in this study. Values between 50% and 60% has the highest frequency. There is a greater number of individuals with lower LVEF values compared to those with higher values creating a left-skewed distribution.

Abbreviations: LVEF, left ventricular ejection fraction

Table 1.

Subject characteristics, surgical features, and complications.

Variable	Mean	SD
Age (yr)	64	11
Weight (kg)	88	19
Height (m)	171	10
Ejection fraction (%)	48	11
Perfusion time (min)	84	39
Cross-damp lime (min)	51	24
Diseased vessels	2.7	0.6
Distal arterial grafts	1.5	0.8
Distal vein grafts	1.8	1
Total grafts	3.2	1
Revascularization index	0.6	0.8
	N	%
Male	5,463	69
Race
Black	366	5
White	6.100	77
Other	215	3
Unknown	1,251	16
Payor
Medicaid	414	5
Medicare	3.987	50
Private	3.312	42
Uninsured	121	2
Unknown	98	1
Smoker	4.940	62
Diabetes mellitus	3.080	39
Hyper lipidemia	5.806	73
Dialysis	133	2
Hypertension	6.644	84
Stroke	662	8
Cerebrovascular disease	2045	26
Chronic I.ung Disease
None	6,245	79
Mild	1,179	15
Moderate	364	5
Severe	144	2
Peripheral vascular disease	1,431	18
Prior cardiac intervention	2571	32
Prior CABG	384	5
Prior valve and other	97	1
ca rduc surgery
Myocardial infarction	4703	59
Congestive heart failure	1.042	13
Angina pectoris	5.827	73
Arrhythmia
Yes	1.089	14
No	6,546	83
Unknown	297	4
NYHA
Class 1	409	5
Class 11	1,066	13
Class III	2,971	37
Class IV	2,423	31
Unknown	1,063	13
Beta blocker use	5,588	70
Aspjrin use	5,921	75
Intravenous Nitrate use	2,112	27
Anticoagulation	3,287	41
Steroids	206	3
Status
Elective	2,626	33
Urgent	4,768	60
Emergent	512	6
Emergent Salvage	26	0.3
Intraaortic balloon pump	882	11
Cardioplegia	7,729	97
Intemal Mammary artery
Both	390	5
Left only	6,743	85
Right only	150	2
None	649	8
Blood transfusion	2,770	35

Abbreviations: CABG, coronary artery bypass grafting NYHA, New York Heart Association: SD, standard deviation.

The characteristics of subjects with preoperative LVEF >65% compared to those with values 55–65% are summarised in Supplemental Table 1. Individuals with LVEF >65% were more likely to be female (50% v. 35%, p<0.001), shorter (1.68±0.1 m v. 1.71±0.1 m, p=0.001) and were more likely to have cerebrovascular disease (35% v. 25%, p=0.016). Surgically, they were more likely to receive venous (1.9±1.2 v. 1.7±1.0, p=0.019) rather than arterial (1.3±0.8 v. 1.5±0.8, p=0.007) grafts.

Overall, 2,810 (35%) of subjects died by late follow-up. Operative mortality occurred in 127 (2%) individuals and complications occurred in 3,008 (38%). The most common complication was new onset atrial fibrillation (n=1,707, 22%) followed by prolonged mechanical ventilation (n=792, 10%) and renal failure (n=406, 5%). The remainder of the postoperative complications are summarised in Supplemental Table 2.

Many patient and surgical factors differed between those who did and did not suffer postoperative complications or mortality. Supplemental Tables 3–6 summarise these differences for late mortality, operative mortality, any complication, and atrial fibrillation, respectively. Across all outcomes, those experiencing these postoperative events were older, had a lower preoperative LVEF, and tended to have a higher prevalence of various comorbidities.

The risk of late and operative mortality, and composite complications appeared to decrease as LVEF increased, but, above certain levels, the risk increased as LVEF further increased. This J-shaped relationship was not present for atrial fibrillation, the risk of which appeared to continuously decrease as LVEF increased (Figure 2). Given these relationships, LVEF values were transformed using fractional polynomial equations for analysis in subsequent models. After adjustment for other factors, transformed LVEF remained associated with operative complications and atrial fibrillation, but not early mortality (Table 2).

Figure 2.

Bubble plots depicting the incidence of postoperative complications for various preoperative LVEF values. The size of each bubble corresponds to the number of individuals with each LVEF value and the dashed line provides the best fit of the relationship between each adverse event over the range of preoperative LVEF values. Except for atrial fibrillation, all complications exhibit a J-shaped relationship, in which both high and low values are associated with an increased incidence of the complication, with a greater increase for lower values compared to higher values. Odds ratio and 95% confidence intervals for the transformed LVEF were calculated using (0.011861354+ 0.312297056*(LVEF)-2 + 0.00002748*(LVEF)-1 + 0.001904649*(LVEF)-0.5+ 0.699678835*LN(LVEF) + 0.000467464 *(LVEF)0.5 + 0.000135754*(LVEF) + 0.031673709 *(LVEF)2)/(0.396569513*(LVEF)-2 + 0.312782862*(LVEF)-1 + 0.000526557*(LVEF)-1 + 0.007870875 * (LVEF)) as the fractional polynomial for composite complications. Odds ratio and 95% confidence intervals for the transformed LVEF using (0.072945099*(LVEF)-2 + 0.02476404*(LVEF)-1 + 0.033256555*(LVEF)-0.5 + 0.060879112*LN(LVEF) + 0.005060298 *(LVEF)0.5 + 0.021605609*(LVEF) + 0.064786727 *(LVEF)2)/(0.159609753 + 0.242193496*(LVEF)-2 + 0.167915547*LVEF)-1 + 0.142360686*(LVEF)-0.5 + 0.095692745*LN(LVEF) + 0.137228665*(LVEF)0.5 + 0.128700597*(LVEF) + 0.217734774 * (LVEF)2) as the fractional polynomial for atrial fibrillation. Odds ratio and 95% confidence intervals for the transformed LVEF using (0.000005045216 + 0.004472213*(LVEF)-2 + 0.005220405*(LVEF)-1 + 0.00031699*(LVEF)-0.5 + 0.038423131*LN(LVEF) + 0.000265586 *(LVEF)0.5 + 0.000114651*(LVEF) + 0.000128124 *(LVEF)2)/(0.069723365*(LVEF)-2 + 0.385526084*LN(LVEF) + 0.03014384*(LVEF)0.5 + 0.0119135*(LVEF) + 0.268748256 *(LVEF)2) as the fractional polynomial for operative mortality.

Abbreviations: LVEF, left ventricular ejection fraction

Table 2.

Association of preoperative LVEF and postoperative complications from multivariable logistic regression models.

Complication	OR	Lower Cl	Upper Cl	p-value
Composite	7.085	3.027	16.58	<0.001
Atrial Fibrillation	75.02	2.830	1988	0.010
Operative Mortality	0.192	0	438412	0.816

Abbreviations: Cl, confidence interval; LVEF, left ventricular ejection fraction, LN, natural logarithm.

The details of the Cox model of late mortality are summarised in Table 3. After adjusting for other preoperative factors, intraoperative factors, and postoperative complications, we found that transformed LVEF was significantly associated with late mortality, hazard ratio=9.223 (95% CI=4.620–18.41, p<0.001) (Table 3). After back transformation, patients with LVEF of 60% had the lowest risk of long-term death (26.7%). Mortality increased as LVEF both increased and decreased from LVEF of 60% (Figure 3). The long-term mortality risk (34.6% or hazard ratio=1.209) for LVEF of 70% was equal to that associated with an LVEF of 51%, while LVEF of 75% (probability of being dead=38.7%, hazard ratio=1.294) was equivalent to an LVEF of 48% (Figure 4).

Table 3.

Cox regression for late mortality.

Variable	Odds Ratio	Lower 95% Cl	Upper 95% Cl	p-value
Age	1.044	1.038	1.050	<0.001
Payor - Private Insurance	1			<0.001
Medicaid	0.781	0.662	0.920	0.003
Medicare	1.371	1.123	1.673	0.002
Uninsured	0.902	0.760	1.070	0.238
Unknown	0.628	0.435	0.907	0.013
Smoker	1.240	1.136	1.353	<0.001
Diabetes melhtus	1.378	1.274	1.491	<0.001
Renal Failure - None	1			<0.001
Unknown	1.258	1.044	1.517	0.016
Yes	1.772	1.449	2.167	<0.001
Dialysis	2.040	1.549	2.685	<0.001
Cerebrovascular Disease	1.246	1.148	1.353	<0.001
Chronic Lung Disease - No	1			<0.001
Mild	1.267	1.144	1.402	<0.001
Moderate	1.410	1.201	1.655	<0.001
Severe	1.802	1.420	2.287	<0.001
Penpheral Vascular Disease	1.298	1.183	1.423	<0.001
Congestive Heart Failure	1.149	1.035	1.276	0.009
Anticoagulant use	0.906	0.833	0.986	0.022
Steroid use	1.524	1.255	1.850	<0.001
Aspirin use	0.897	0.824	0.976	0.012
Diseased vessel - One	1			0.017
Two	0.873	0.763	0.998	0.047
Three	1.131	1.039	1.233	0.005
Status - elective	1			0.007
Urgent	1.171	0.989	1.386	0.067
Emergent ♦ Emergent Salvage	1.147	1.051	1.251	0.002
Perfusion time (min)	1.007	l.OOS	1.008	<0.001
Cross clamp lime (min)	0.994	0.991	0.997	<0.001
Number distal arterial grafts	0.797	0.736	0.861	<0.001
Number distal venous grafts	0.871	0.824	0.921	<0.001
Left Internal mammary artery use				0.003
Both	0.746	0.572	0.972	0.030
None	0855	0.739	0.989	0.035
Right	0.708	0.537	0.932	0.014
Complications
Redo graft	0.284	0.126	0.643	0.003
Reoperation for other cardiac reasons	1.502	1.197	1.884	<0.001
Reoperation for non-cardiac reasons	1.256	1.034	1.524	0.021
Myocardial infarction	1.729	1.312	2.278	<0.001
Thoracotomy site infection	1351	1.883	96.93	0.010
Stroke	1.378	1.065	1.783	0.015
Coma	1.667	1.170	2.375	0.005
Pneumonia	0.659	0.521	0.834	<0.001
Renal failure	1.259	1.081	1.466	0.003
Iliac or femoral artery dissection	0.137	0.018	1.056	0.056
Acute limb ischemia	3.015	1.840	4.940	<0.001
Arrest	5.497	4.414	6.846	<0.001
Coagulation	1.991	1.217	3.258	0.006
Multisystem organ failure	1.407	1.069	1.851	0.015
Alnal fibrillation	1.097	1.003	1.199	0.043
Aortic dissection	13.08	4.587	37.30	<0.001
Ejection fraction - transformed	9.223	4.620	18.41	<0.001

Abbreviations: Cl. confidctvc interv al, LVEF, left ventricular qcction fraction.

Figure 3.

Display of the hazard ratio for late mortality for individuals with different preoperative LVEF values. The hazard was lowest for individuals with an LVEF of 58%. Risk increased for values both above and below this value. Hazard ratio and 95% confidence intervals for the transformed LVEF using (0.125426066*(LVEF)⁻² + 0.682861785*LN(LVEF) + 0.000470897 *(LVEF)^0.5 + 0.000131582*(LVEF) + 0.032823261 *(LVEF)2)/(0.00000305347+ 0.282713505*(LVEF)⁻² + 0.07507182*(LVEF)⁻¹ + 0.00000474722*(LVEF)^−0.5 + 1.70876027*LN(LVEF) + 0.00000440787*(LVEF)^0.5 + 0.010557596*(LVEF) + 0.00000689286 * (LVEF)²) as the fractional polynomial, where LVEF=left ventricular ejection fraction(LVEF), LN=natural logarithm.

Abbreviations: LVEF, left ventricular ejection fraction; LN, natural logarithm

Figure 4.

Survival curves for different levels of preoperative LVEF values demonstrating the negative impact of increased preoperative values on long-term survival. Those with a preoperative LVEF=70% has worse survival than those with an LVEF=60%, but better than those with severely reduced values.

Abbreviations: LVEF, left ventricular ejection fraction

No association existed between transformed LVEF and operative mortality after adjusting for other preoperative and intraoperative factors (Supplemental Table 7). When the transformed LVEF variable was forced into the model, it had an OR of 0.196 (95% CI 0–459604, p=0.827).

Regarding the composite complication outcome, after adjusting for preoperative and intraoperative factors, the transformed LVEF had a significant association with the risk of developing any complication being lowest for those with a LVEF of 64% (33% risk of complications) (Supplemental Table 8). The LVEF either above or below this value had increased risk. Those with an LVEF of 75% had a 34% risk while those with an LVEF of 80% had a 36% risk (Figure 2). Similar risk was found for decreased LVEF values of 54% and 49% respectively. Risk continued to increase as LVEF decreased with a preoperative LVEF of 15% being associated with a 64% increased risk.

The occurrences of postoperative atrial fibrillation also remained associated with the transformed LVEF variable after adjustment for other factors (Supplemental Table 9). This outcome did not demonstrate a J-shaped relationship. Instead, the highest rate of atrial fibrillation occurred at the lowest LVEF and the rate progressively decreased as LVEF increased (Figure 2).

Using mediation analysis to determine the direct and indirect effects of LVEF on late mortality, we found that the indirect effect was 7.085 (odds ratio of transformed LVEF on complications) × 1.212 (hazard ratio of complications on mortality) = 8.588. As the total effect was 12.91 (hazard ratio of transformed LVEF on mortality with complications excluded from that model—Table 10), the direct effect is 12.91—8.588=4.32. Hence, two-thirds of the effect of LVEF on late mortality is mediated through its associations with postoperative complications leading to death and only one-third via direct effects on mortality.

Discussion

By accounting for potential nonlinear relationships, we found that a J-shaped curve best describes the relationship between LVEF and multiple adverse events after CABG. This was accomplished by finding the fractional polynomial equation that best fit the data. Specifically, these results suggest that, in addition to lower LVEF values, higher LVEF values may also be associated with increased late mortality, operative mortality, and composite complications. While individuals with higher preoperative LVEF values represented a minority of the overall study population, these findings are significant given the lack of studies addressing the implications of this preoperative finding.

Common scoring systems for calculating the preoperative risk of patients undergoing cardiac surgery include the Society of Thoracic Surgeons (STS) score [16] and EuroSCORE [21]. Both place the same level of risk on all LVEF >50%. Our findings of possible increased risk of the composite complication outcome and late mortality in patients with higher preoperative LVEF models is novel and would be missed using the strategy of common scoring systems. This was accomplished by using fractional polynomials, which make no assumptions about symmetry of the associations or skewness of the data, to determine the best fitting functional form of the relationships between LVEF and the outcomes. By limiting the numbers of terms in the fractional polynomial, we produce a good fit with less risk of overfitting. If confirmed by other studies, our findings suggest that risk models should include higher values of preoperative LVEF as a risk factor for late mortality as well as for composite complications. These future studies will also need to further investigate operative mortality. Since we had only 127 operative deaths, we might have been underpowered to find an association between LVEF and this outcome.

While multiple studies have demonstrated that reduced LVEF places patients undergoing CABG surgery at higher risk [22–24], to our knowledge, our study is unique in determining the relationship between higher LVEF values and mortality and complications in this population. Previous studies of cardiac surgery patients categorise all values above a certain proportion within the normal range or use linear equations to model the relationship [18–20]. By using fractional polynomials to model the relationship, we made no a priori assumptions about the shape of the curve and, importantly, the association of LVEF with adverse outcomes. Our results suggest that there is an optimal LVEF and values above or below this value are associated with worse outcomes. We found the optimal value to be approximately 60%. While the association between increased LVEF and adverse events after CABG surgery is novel, it has been described in other settings involving an acute physiologic insult. For example, Paonessa et al. demonstrated that hyperdynamic LVEF in septic patients was associated with a 38% increased odds of 28-day mortality compared to those with normal LVEF, suggesting that higher LVEF values are important in multiple clinical settings [25].

The mechanisms driving increased incidence of death and complications are not currently known, but must be investigated given their presence in cardiac surgery, non-cardiac surgery [13], and the general population [12]. Diastolic dysfunction likely contributes to this relationship. This would be consistent with previous studies, such as that by Metkus et al., who described a positive association between diastolic dysfunction and multiple adverse events after cardiac surgery that was independent of systolic function [26]. It would also be consistent with the observation by Wehner et al. that a low end-systolic volume index (<10 mL/m²) contributed to this relationship [12].

Other factors could contribute to the relationship between higher LVEF values and adverse events. For example, higher preoperative LVEF could be a marker of conditions that are less chronic than diastolic dysfunction. For example, the presence of sepsis, which reduces afterload, or hypovolaemia, which reduces preload, could both produce an increased preoperative LVEF [26]. It could also arise from increased sympathetic tone representing the presence of more severe preoperative physiologic derangements, which would be consistent with other studies that have shown an association between increased preoperative sympathetic tone and surgical risk [27]. A relationship between increased LVEF values and risk may also exist in postoperative situations where these and other conditions, such as vasoplegia from cardiopulmonary bypass, are present.

While these results offer a new perspective on interpreting LVEF values for patients undergoing cardiac surgery, several limitations should be considered. First, the retrospective study design does not allow for adjustment of unmeasured variables. Also, patients who were excluded from this study because they did not have a preoperative LVEF value differed in regard to race and surgical status when compared to the study population. While the effect of this is likely small given that most patients did have a preoperative LVEF value, this represents another potential source of bias. Second, additional measures of cardiac dimensions and function, such as the end-diastolic volume or measures of diastolic dysfunction were not available to provide insights about the mechanisms driving the observed relationships between LEF and outcomes after CABG. Finally, while the dataset used in this study was selected because it contained precise LVEF measurements instead of the categories seen in many others. However, it had the disadvantage of being older, spanning multiple years, and being restricted to a single institution. Therefore, additional studies across multiple institutions are needed to confirm these findings.

One must also consider that the majority of LVEF measurements were obtained by ventriculography, which may not match findings of other imaging techniques. While ventriculography was previously considered the gold standard for LVEF quantification, cardiac magnetic resonance imaging is current considered the most accurate method for LVEF quantification. Since cardiac catheterisation is frequently performed prior to cardiac surgery, information from this study is frequently available prior to CABG. However, echocardiographic measurements are frequently relied on despite that there is substantial variability between echocardiography and cardiac magnetic resonance imaging [28]. Similar findings have been found in studies comparing echocardiography with ventriculography [29] and these differences may be more pronounced in individuals with coronary artery disease and wall motion abnormalities [30]. Based on this, caution should be taken when extrapolating these findings to LVEF values obtained from different imaging techniques.

Conclusion

Higher LVEF values prior to CABG may represent a distinct phenotype that is at increased risk for postoperative complications and late-mortality. Future studies are needed to confirm these findings and to better characterise this population.